https://axon.cs.byu.edu/~martinez/classes/678/Presentations/Yao.pdf

Extreme learning machine: Theory and applications

I have developed "tennis elbow" from lugging this book around the past four weeks, but it is worth the pain, the effort, and the aspirin. It is also worth the (relatively speaking) bargain price. Including appendixes, this book contains 894 pages of text. The entire panorama of the neural sciences is surveyed and examined, and it is comprehensive in its scope, from genomes to social behaviors. The editors explicitly state that the book is designed as "an introductory text for students of biology, behavior, and medicine," but it is hard to imagine any audience, interested in any fragment of neuroscience at any level of sophistication, that would not enjoy this book. The editors have done a masterful job of weaving together the biologic, the behavioral, and the clinical sciences into a single tapestry in which everyone from the molecular biologist to the practicing psychiatrist can find and appreciate his or

Principles of Neural Science

This paper surveys the field of reinforcement learning from a computer-science perspective. It is written to be accessible to researchers familiar with machine learning. Both the historical basis of the field and a broad selection of current work are summarized. Reinforcement learning is the problem faced by an agent that learns behavior through trial-and-error interactions with a dynamic environment. The work described here has a resemblance to work in psychology, but differs considerably in the details and in the use of the word "reinforcement." The paper discusses central issues of reinforcement learning, including trading off exploration and exploitation, establishing the foundations of the field via Markov decision theory, learning from delayed reinforcement, constructing empirical models to accelerate learning, making use of generalization and hierarchy, and coping with hidden state. It concludes with a survey of some implemented systems and an assessment of the practical utility of current methods for reinforcement learning.

/pdf/reinforcement-learning-a-survey-38lmyylg2y.pdf

Reinforcement learning: a survey

Planning algorithms are impacting technical disciplines and industries around the world, including robotics, computer-aided design, manufacturing, computer graphics, aerospace applications, drug design, and protein folding. This coherent and comprehensive book unifies material from several sources, including robotics, control theory, artificial intelligence, and algorithms. The treatment is centered on robot motion planning but integrates material on planning in discrete spaces. A major part of the book is devoted to planning under uncertainty, including decision theory, Markov decision processes, and information spaces, which are the “configuration spaces” of all sensor-based planning problems. The last part of the book delves into planning under differential constraints that arise when automating the motions of virtually any mechanical system. Developed from courses taught by the author, the book is intended for students, engineers, and researchers in robotics, artificial intelligence, and control theory as well as computer graphics, algorithms, and computational biology.

Planning Algorithms: Introductory Material

This paper surveys the field of reinforcement learning from a computer-science perspective. It is written to be accessible to researchers familiar with machine learning. Both the historical basis of the field and a broad selection of current work are summarized. Reinforcement learning is the problem faced by an agent that learns behavior through trial-and-error interactions with a dynamic environment. The work described here has a resemblance to work in psychology, but differs considerably in the details and in the use of the word ``reinforcement.'' The paper discusses central issues of reinforcement learning, including trading off exploration and exploitation, establishing the foundations of the field via Markov decision theory, learning from delayed reinforcement, constructing empirical models to accelerate learning, making use of generalization and hierarchy, and coping with hidden state. It concludes with a survey of some implemented systems and an assessment of the practical utility of current methods for reinforcement learning.

Reinforcement Learning: A Survey

A digital data acquisition system has been flight tested in a Schweizer 2-32 sailplane and the algorithms, designed for high angle of attack parameter identification, are described in this paper. The system is portable and self-contained. Inertial sensors are mounted in a strapdown unit. Two Kiel-probes are used for airspeed measurement. "Off-the- shelf" microelectronics components are configured to operate as a multiprocessing system. The on-board computing power is used to monitor the sensor and support system and to improve flight test effectiveness. The Estimation-Before-Modeling technique has been selected for parameter identification. A time history of a flight test maneuver that will be used in the estimation of the aerodynamic parameters is also presented.

/pdf/data-acquisition-system-and-methodology-for-high-angle-of-4t5az8j8it.pdf

Data acquisition system and methodology for high angle of attack parameter estimation

A technique for determining the most probable failure state of a restructurable control system is presented. The approach is to build a knowledge base that contains and makes use of inference mechanisms to deduce the most likely failures given the symptoms. The analysis is first carried out in a local sense, where only probabilistic information and causality are used to generate failure models, then in a global sense, where the models are grouped and heuristics are used to prune the number of candidate models. Procedures are illustrated using failure patterns of a generic database as well as a fault scenario for a hypothetical helicopter flight control system. It is concluded that the methods are potentially capable of handling generic failures and thus are useful in truly restructable control systems.

Failure Model Determination in a Knowledge-Based Control System

Robust controllers for nonlinear systems with uncertain parameters can be reliably designed using probabilistic methods. In this chapter, a design approach based on the combination of stochastic robustness and dynamic inversion is presented for general systems that have a feedback-linearizable nominal system. The efficacy of this control approach is illustrated through the design of flight control systems for a hypersonic aircraft and a highly nonlinear, complex aircraft model. The proposed stochastic robust nonlinear control explores the direct design of nonlinear flight control logic; therefore the final design accounts for all significant nonlinearities in the aircraft’s high-fidelity simulation model. Monte Carlo simulation is used to estimate the likelihood of closed-loop system instability and violation of performance requirements subject to variations of the probabilistic system parameters. The stochastic robustness cost function is defined in terms of the probabilities that design criteria will not be satisfied. We use randomized algorithms, in particular genetic algorithms, to search the design parameters of the parameterized controller with feedback linearization structure. The design approach is an extension of earlier methods for probabilistic robust control of linear systems. Prior results are reviewed, and the nonlinear approach is presented.

Probabilistic Control of Nonlinear Uncertain Systems

The objective of this paper is to interpret existing flying qualities criteria in a multivariable format based upon state vectors and matrices of stability and control derivatives. State-space methods are shown to be entirely compatible with conventional flying qualities analysis. Furthermore, they lead to a definition of the well-known control anticipation parameter that has potential utility in the evaluation of higher-order aircraft/control systems.

A Unifying Framework for Longitudinal Flying Qualities Criteria

Treatment of a viral disease process is interpreted as the optimal control of a dynamic system. Evolution of the disease is characterized by a nonlinear, fourth-order ordinary differential equation that describes concentrations of pathogenic antigens (or pathogens), plasma cells, and antibodies, as well as a numerical indication of patient health. The dynamic equations are controlled by therapeutic agents that affect the rate of change of system variables. Control histories that minimize a quadratic cost function are generated by numerical optimization over a fixed time interval, given otherwise lethal initial conditions. Tradeoffs between cost function weighting of pathogens, organ health, and use of therapeutics are evaluated. Optimal control solutions that defeat the virus and preserve organ health are demonstrated for individual and combined therapies. It is shown that control theory can point the way toward new protocols for treatment and cure of human diseases.

Robert F. Stengel

Papers

Data acquisition system and methodology for high angle of attack parameter estimation

Failure Model Determination in a Knowledge-Based Control System

Probabilistic Control of Nonlinear Uncertain Systems

A Unifying Framework for Longitudinal Flying Qualities Criteria

Optimal control of a viral disease